THE EXACT PENALTY-FUNCTION METHOD IN CONSTRAINED OPTIMAL-CONTROL PROBLEMS

This paper uses the exact penalty function method to transform a const rained optimal control problem into an unconstrained one and establish es an equivalence between the two problems in the ''local'' sense. Nec essary and sufficient condi tions are obtained for a penalty function to be exact. This generalizes the result of Xing et al. (J. Optim. Con trol Appl. Methods 10(2) (1989), 173-180) where only sufficient condit ions are obtained. As a by-product, the relationship between a penalty function and a stationary point is also established. (C) 1994 Academi c Press, Inc.